JP6549538B2 - Congestion control apparatus and congestion control method - Google Patents

Description

  The present invention relates to a technique for controlling congestion in a server of a virtualization configuration.

  A voice communication server that performs call control processing incorporates congestion control for preventing performance degradation due to lack of resources when server load increases due to congestion. By quickly detecting congestion and temporarily restricting the reception of new calls, it is possible to calm down congestion and avoid large-scale failures.

  Conventionally, a voice communication server is configured by operating one application on one hardware. As shown in FIG. 5, in congestion control in a configuration in which one application is operated for one hardware, the upper limit value of the CPU utilization is set based on the limit value of the hardware performance, and the CPU utilization is When the upper limit is reached or approached, restriction control is performed to discard the reception of a new call.

  In recent years, with the penetration of virtualization technology, there is a tendency to operate multiple applications on one hardware.

Muraaki Masaaki, 3 others, "Application of virtualization technology to communication service", Magazine FUJITSU, Fujitsu Limited, November 2012, Vol. 63, No. 6, pp. 640-645

  As congestion control in a virtualization configuration in which multiple applications (virtual servers) are operated on one hardware (physical server), a method of managing resource consumption in units of virtual servers, and total resource consumption of virtual servers Can be managed.

  In the method of managing resource consumption on a per virtual server basis, as shown in FIG. 6, individual hardware performance limits are set for each of the virtual servers A and B. Then, each of the virtual servers A and B determines whether the CPU utilization has reached the hardware performance limit, and the virtual server A having the CPU utilization reaching the hardware performance limit performs restriction control and performs a new call. Do not receive The virtual server B whose CPU utilization has not reached the hardware performance limit does not perform restriction control.

  In the method of totalizing and managing the resource consumption of the virtual server, as shown in FIG. 7, a plurality of virtual servers A and B share the hardware performance limit. It is determined whether the value obtained by totaling the CPU utilization of each virtual server has reached the hardware performance limit, and if the total value of the CPU utilization has reached the hardware performance limit, any virtual server is also subjected to restriction control Do not receive new calls.

  However, each of the methods described above has a problem.

  In the method of managing resource consumption on a per virtual server basis, although there is a surplus of resources of the physical server, the virtual server A is restricted from receiving new calls. Therefore, there is a problem that it is not possible to make the best use of the total performance of physical servers.

  In the method of managing the sum of resource consumption of virtual servers, when the CPU utilization of one virtual server A increases, the upper limit of resources that can be used by the other virtual server B decreases, and the virtual server A and B have a lower upper limit. There was a problem that the balance was lost.

  The present invention has been made in view of the above, and it is an object of the present invention to perform congestion control in a server of a virtualization configuration in consideration of effective utilization of resources and balance of resources.

  A congestion control apparatus according to a first aspect of the present invention is a congestion control apparatus for controlling congestion of a plurality of virtual servers operating in a virtualization configuration, and obtaining means for obtaining resource usage of each of the plurality of virtual servers The number of new calls received by each of the plurality of virtual servers based on the resource usage of each of the plurality of virtual servers when the total value of the resource usage of the plurality of virtual servers reaches a first threshold A second restriction unit that restricts each of the plurality of virtual servers not to receive a new call when the total value of the resource usage of the plurality of virtual servers reaches a second threshold. And means.

  A congestion control method according to a second aspect of the present invention is a congestion control method for controlling congestion of a plurality of virtual servers operating in a virtualization configuration, and acquiring resource usage of each of the plurality of virtual servers; When the total value of the resource usage of the plurality of virtual servers reaches a first threshold, the number of new calls received by each of the plurality of virtual servers is determined based on the resource usage of each of the plurality of virtual servers Providing a restricting step, and restricting each of the plurality of virtual servers not to receive a new call when the total value of resource usage of the plurality of virtual servers reaches a second threshold. It features.

  According to the present invention, congestion control can be performed in consideration of effective utilization of resources and balance of resources in a server having a virtualization configuration.

It is a functional block diagram which shows the structure of the congestion control apparatus in this Embodiment. It is a flowchart which shows the flow of a process of the congestion control apparatus in this Embodiment. It is a figure which shows CPU utilization of a virtual server when the total value of CPU utilization reaches a threshold value. It is a figure which shows a mode that the congestion control apparatus in this Embodiment controls. It is a figure explaining congestion control in composition which operates one application to one conventional hardware. It is a figure explaining congestion control which manages resource consumption per virtual server. It is a figure explaining the congestion control which totals and manages the resource consumption of a virtual server.

  Hereinafter, embodiments of the present invention will be described using the drawings.

  FIG. 1 is a functional block diagram showing a configuration of a congestion control device in the present embodiment. The congestion control device 1 illustrated in FIG. 1 includes a virtualization unit 11, an acquisition unit 12, and a regulation unit 13.

  The virtualization unit 11 operates a plurality of virtual servers A and B by sharing resources of one hardware (physical server) using a virtualization technology. The virtual servers A and B according to the present embodiment are call processing servers that receive new calls and execute predetermined call processing. Although only two virtual servers A and B are illustrated in FIG. 1, the number of virtual servers A and B operating on the virtualization unit 11 is not limited to two.

  The acquisition unit 12 acquires the consumption of resources (CPU, memory, etc.) used by each virtual server A, B operating in the virtualization unit 11 or the load of each virtual server A, B.

  When the resource consumption of the virtual servers A and B satisfies a predetermined condition, the restricting unit 13 restricts reception of new calls of the virtual servers A and B based on the resource consumption of the virtual servers A and B. . Specifically, the restriction is not activated when the total value of the resource consumption of the virtual servers A and B does not reach the predetermined threshold. When the total value of resource consumption of virtual servers A and B reaches a predetermined threshold, the degree of regulation is adjusted based on the ratio of resource consumption of virtual servers A and B, and the performance limit of physical servers is reached. The virtual servers A and B regulate the number of new calls to be received so that resource consumption can be balanced between the virtual servers A and B as they approach (1). If the total value of resource consumption of virtual servers A and B reaches the performance limit of the physical server, all virtual servers A and B restrict reception of new calls and do not accept new calls (secondary restriction) . The predetermined threshold is set lower than the performance limit of the physical server.

  Next, the flow of processing of the congestion control device in the present embodiment will be described.

  FIG. 2 is a flowchart showing a flow of processing of the congestion control device in the present embodiment.

  The acquisition unit 12 acquires the CPU usage rate (resource consumption amount) of each virtual server A, B, and the restriction unit 13 determines whether the total value of the CPU usage rates of the virtual servers A, B has reached a threshold value Is determined (step S11). If the total value of the CPU utilization does not reach the threshold value, no restriction is made (No in step S11). Note that another determination method may be used as the determination method of step S11. For example, it is determined whether the ratio of the CPU utilization of the virtual servers A and B exceeds a predetermined ratio. Alternatively, it is determined whether the CPU utilization of either of the virtual servers A and B has reached a predetermined value (for example, 50%).

  When the total value of the CPU utilization reaches the threshold (Yes in step S11), the restriction unit 13 starts primary restriction so that the CPU utilization can be balanced between the virtual servers A and B (step S12).

  A specific control method will be described using FIGS. FIG. 3 is a diagram showing the CPU utilization of the virtual servers A and B when the total value of the CPU utilization reaches a threshold (“regulatory operation triggering threshold” in the drawing). As shown in FIG. 3, it is assumed that the ratio of the CPU utilization of virtual server A to the CPU utilization of virtual server B is M: N. In this case, as shown in FIG. 4, the restriction unit 13 sets the ratio of the number of new calls received by virtual server A to the number of new calls received by virtual server B per unit time to N: M, that is, the reciprocal of the CPU utilization Regulate the number of new calls received so that the ratio of As a result, the ratio of the CPU utilization of the virtual servers A and B can be balanced as the performance limit is approached. In addition, the control method should just be able to balance CPU utilization of the virtual servers A and B.

  After starting the primary restriction, the restriction unit 13 determines whether the total value of the CPU utilization of the virtual servers A and B has reached the performance limit (step S13). If the total value of the CPU utilization has not reached the performance limit, the primary regulation is continued (No in step S13).

  If the total value of the CPU utilization reaches the performance limit (Yes in step S13), the restricting unit 13 restricts the reception of new calls by the virtual servers A and B, and starts secondary restriction not accepting new calls. (Step S14).

  After that, when the congestion state disappears and the total value of CPU utilization does not reach the performance limit, the restriction is gradually released. Specifically, if the total value of the CPU utilization has not reached the performance limit, but is equal to or greater than a predetermined threshold, the secondary regulation is changed to the primary regulation. There is no restriction when the total value of CPU utilization does not reach a predetermined threshold.

  As described above, according to the present embodiment, when the total value of the CPU utilization of virtual servers A and B operating on virtualization unit 11 reaches a predetermined threshold value, virtual server A, B regulates the number of new calls received so that the ratio of the number of new calls received by each unit time per unit time becomes the ratio of the reciprocal of the resource usage of each of the virtual servers A and B, and operates on the virtualization unit 11 If the total value of CPU utilization of virtual servers A and B reaches the performance limit of the physical server, the CPU utilization of virtual servers A and B is balanced by restricting reception of new calls. In addition, resources of physical servers can be used effectively.

1 Congestion control apparatus 11 Virtualization unit 12 Acquisition unit 13 Restriction unit

Claims (4)

A congestion control apparatus for controlling congestion of a plurality of virtual servers operating in a virtualization configuration, comprising:
Acquisition means for acquiring the resource usage of each of the plurality of virtual servers;
When the total value of the resource usage of the plurality of virtual servers reaches a first threshold, the number of new calls received by each of the plurality of virtual servers is determined based on the resource usage of each of the plurality of virtual servers First regulation means to regulate,
Second restricting means for restricting each of the plurality of virtual servers not to receive a new call when the total value of resource usage of the plurality of virtual servers reaches a second threshold;
A congestion control apparatus characterized by having:   The plurality of first restriction means are configured such that a ratio of the number of new calls received by each of the plurality of virtual servers per unit time is a ratio of inverse numbers of resource usage of the plurality of virtual servers. The congestion control device according to claim 1, wherein the number of new calls received by the virtual server is restricted. A congestion control method for controlling congestion of a plurality of virtual servers operating in a virtualization configuration, comprising:
Acquiring resource usage of each of the plurality of virtual servers;
When the total value of the resource usage of the plurality of virtual servers reaches a first threshold, the number of new calls received by each of the plurality of virtual servers is determined based on the resource usage of each of the plurality of virtual servers Steps to regulate,
Restricting each of the plurality of virtual servers not to receive a new call when the total value of resource usage of the plurality of virtual servers reaches a second threshold;
A congestion control method characterized by comprising:   The step of restricting the number of new calls received may be a ratio of the number of new calls received by each of the plurality of virtual servers per unit time to a ratio of inverse numbers of resource usage of each of the plurality of virtual servers. 4. The congestion control method according to claim 3, wherein the number of new calls received by the plurality of virtual servers is restricted to be.

Images